How can I use containers to reduce cloud costs?

Although the mainstream of several major public cloud providers have achieved beyond imagined efficiency, however, at any given time, they still have idle excess capacity. AWS and Google Compute Engine are willing to sell these resources at very low discounts in order to make these resources available and get some returns. The discount is usually 90%.

What is this insider? Prices are market-driven and set by the highest bidder. This is a typical market model: asset value driven by market demand. However, for public cloud users, the challenge is that at any given time, if someone bids over you, the spot instance you use will be reclaimed. Before Amazon, you have two minutes to clear the instance before the cloud service terminates, and the same time Google Cloud gives you 30 seconds.

This fluctuation makes most companies using public clouds remain cautious about this model. If the user at any time may lose the server, especially in setting the server ready to spend a long time, how to keep the user's application running? It is not uncommon for configuration management tools to take 10 minutes or more to install the package and deploy the application. The length of time required to set up the server, plus the critical period of time, makes it very challenging to use these discount instance types effectively.

How the container helps optimize cloud costs

You may have guessed that by using the spot market, the container can help you solve this problem well. The predictability of a container means that the start-up time can be greatly reduced compared to traditional dynamic, scripted, or configuration-managed methods. The required packages, application code, and various files are ready for build and are written to a compressed archive (Docker image). This means that the application startup time is within a minute, is already possible.

In addition, the container allows you to deploy the application to the new host at ease, and it is sure that it will work as expected. An independent dependency model means that everything that the application needs will run through the whole process. You do not have to worry that your automation tool will fail in one of the required packages and eventually become a bad node.

Optimize cloud costs with Rancher and Spotinst

In order to further improve the stability of the use of field examples, you can use some of the outstanding products such as Spotinst Elastigroup. Spotinst Elastigroup uses predictive algorithms to help you anticipate market behavior and to migrate workloads when the market-driven spot price is higher than the list in different spot types (based on price and availability) and on-demand equivalents. Spotinst as a forecasting layer will ensure you get the best possible cost to meet your needs.

With Spotinst, you only need to create a pool of instance types that apply to your host. Spotinst will choose which services to offer based on current prices and market stability. What you have to do is define the instance type you want to use as the host. Since Spotinst is cloud-independent, you can define individual Elastigroups in AWS, GCP, and Azure, and use the Spotinst API to scale for your liking. (For more information on Elastigroups how to get the spot market to help you optimize your costs, please click here ).

Spotinst provides long-term integration with Rancher and automatically adds replacement nodes to the Rancher cluster so that containers can be migrated on containers that are used for replacement. Spotinst will instruct Rancher to suspend each interrupted container and relocate it to another instance. Integration with Rancher makes it easier to use the spot market, without sacrificing the performance of the application.

understand more

To gain the full value of the container, such as increasing developer agility, simplifying CI / CD workflows, and achieving better resiliency, you can join Rancher's official technical exchange group to interact with more fellow people.

Source: Rancher Labs

Heads up! This alert needs your attention, but it's not super important.